33th Annual North American Symposium on Bat Research, October 2003

Computer Vision for Tracking Bats in Infrared Thermal Video: A Tool for Understanding the Behavior of Bats in Flight

Boston University

D. E. Hirsh, M. Betke, S. Crampton, J. Horn and T. H. Kunz

The ability to observe and automatically follow the paths of individual bats in video sequences would allow better understanding of the flight patterns of bats while emerging, commuting, foraging, and returning. Additionally, this ability would allow computational analysis of the differences in flight patterns observed in different species, or at different stages of development. We recorded the nightly activity of Tadarida brasiliensis with a Merlin mid infrared thermal camera and developed a method to follow the flight paths of individual bats appearing in the video. Our method builds upon previous work to automatically analyze infrared thermal video. It utilizes predictive search and geometrical conflict resolution techniques. The paths of individual bats can be followed when individuals change direction and speed, and when individuals appear in moderately dense groups. This tracking method promises to expand our ability to better understand the movement, interaction, and group behavior of bats in flight.